AGENTYK
Cloud platform

AgentykCloud

Privacy-first AI chat in your browser. Stateless server, broad model selection, and no conversation history kept on our side.

Where to use

What it does

Stateless server

Your messages flow through the cloud only long enough to reach the model. We don't keep a database of conversations.

Conversations live in your browser

Chat history is stored in your browser's localStorage with auto-prune. Clearing the browser clears your history.

Frontier-tier open models

The full AgentykLM lineup, Hummingbird through Grizzly, selectable per chat — including the 256K-context long-document tier. Open weights, EU-hosted inference; the Orca tier is on the way.

100% Dutch hosted

Inference and edge servers run in EU jurisdiction under GDPR. No CLOUD Act exposure, no data leaves Europe.

In-process tools

Calculator, date / time, unit conversion, and document Q&A run inside the chat process — no third-party calls.

Web search opt-in

Need current info? Toggle web search per chat. Your query goes to Mojeek (independent EU search) — clearly marked in the UI.

Audio to text with Agentyk Scribe

Record your voice or drop an audio file right in the chat and Agentyk Scribe transcribes it to text in place — then chat with, summarise, or translate the transcript. EU-hosted; the audio is never stored.

Made for

Companies handling sensitive information

Legal, healthcare, finance, defense — that need EU jurisdiction and verifiable data handling instead of a “we don't train on your data, trust us” promise from a US-based provider.

Journalists and researchers

Protecting sources, drafts, interview transcripts, and unpublished findings from third-party AI vendors.

Educational institutions

With student-data privacy obligations under GDPR who need an AI that doesn't ship usage data abroad.

Public sector and NGO staff

Handling constituent records, casework, or confidential dossiers under data-residency rules.

Knowledge workers

Who want a capable AI assistant without their conversations being kept on a database somewhere or mined for training data.